LIMITS OF ERROR 53 



to the Government of the day. The salary was then 

 doubled, " thirty chaldrons of coals and one hundred 

 pounds of wax candles" were asked for, and the 

 enclosing of the kitchen - garden ! Evidently the 

 official mind had not grasped the idea that the 

 astronomer - royal was no longer a fortune- telling 

 interpreter of the heavens, as Kepler had been forced 

 to become for bread ! With one assistant all - night 

 work was barely possible ! 1 The instruments in use 

 may be judged of from " An Account of the Equatorial 

 Instrument," or " mural quadrant," given to the Royal 

 Society in 1793, twenty years after Herschel began 

 his labours. The precision of observation among 

 the ancients could not be trusted to within from five 

 to ten minutes. Tycho Brahe reduced the probable 

 limit of error to within one minute. Hevelius in the 

 following century brought it down to fiften or twenty 

 seconds, and in the century after it was reduced to seven 

 or eight seconds. 2 To entitle observations to any 

 credit it was then felt that a probable error of more 

 than a few seconds could not be admitted or perhaps 

 only a hundredth part of the errors unavoidable in 

 the days of Hipparchus. In 1827, Sir John Herschel 

 was able to say that he had " secured such a degree of 

 precision that the stars cross the wire often on the 

 very beat of the chronometer when they are expected." 

 Clocks, transit-instruments, mural-circles may be said 

 to have been in their infancy when Herschel began 

 his work. He did not propose to work or measure 

 with these as men do in an observatory. He was 

 eager to see with a telescope; but he soon found 

 that, if he was to do any good, he would require to 

 1 Weld, ii. 250. a Phil. Trans, for 1793. 



